Skip to content

Pricing

Pay for classifications, not for months.

There is no subscription and no catalog limit. You are charged for each product DutyLookout classifies — the first 25 every month are free. Classify your catalog once, and after that you only pay for the products you add.

Rates

Rates are graduated, so each band applies only to the classifications inside it — like tax brackets. Classifying more never costs you more per product.

Classifications per monthPrice each
First 25Free
26–100$0.09
101–1,000$0.055
1,001–5,000$0.035
5,001+$0.032

What that actually costs

The first month covers your existing catalog. After that you only pay for new products, so the ongoing cost is a fraction of a subscription.

CatalogFirst monthThen, adding ~2% a month
Small catalog100 products$6.75Free — under the 25/month allowance
Growing store1,000 products$56.25Free — under the 25/month allowance
Large catalog5,000 products$196.25$6.75 / month

Prices in USD, billed through Shopify on your regular app-billing invoice. Taxes may be added by Shopify depending on your region.

What the price buys

There are cheaper ways to put an HS code on a product, and at some catalog sizes there are cheaper ways to buy this one. What a classification costs here reflects the work done on it — a second verification pass, a frontier model, and the monitoring afterwards all cost us money per product, every time.

Two passes

A second pass against the real tariff lines

One pass proposes the heading and subheading. A separate second pass then reads the actual published 10-digit statistical lines that exist underneath it and chooses among them. That matters because the first pass never sees which lines are real — it is answering from memory, and a plausible code under the right subheading is the failure mode nothing downstream can catch.

Real codes

Nothing we hand back is invented

Every code is resolved against the tariff revision currently published, so it is a line that genuinely exists rather than one that merely looks well-formed. Where an 8-digit prefix has more than one statistical line under it, we refuse to pick for you — the product is flagged for a person instead of being guessed at.

Model

A frontier model on the paid path

Merchant classification runs on Claude Sonnet 5. The free lookup on this site runs on the smaller Claude Haiku 4.5 — which is the difference between an indicative answer and one you are prepared to put on a customs entry. Frontier-model tokens are a real per-classification cost, and they are part of what the per-product price is for.

Your codes

Codes you already have are tried first

If a product already carries a code, yours is attempted before anything is generated, and kept when it checks out against the published schedule. When it does not, the product is flagged for a human rather than quietly replaced. You are not paying us to overwrite work you already did.

Arithmetic

No model anywhere in the numbers

Duty is computed, not predicted. It is ordinary arithmetic against a versioned rate snapshot that carries its revision label, with Section 301 and 232 and the other surcharge programs resolved by effective-date window at a stated moment in time. Every estimate names the snapshot it came from. The model chooses the code; it never touches the money.

Watch

Monitoring is in the price, not an add-on

A nightly job diffs the new rate snapshot against the one you were quoted from, and separately checks whether a surcharge window opened or closed overnight — a change that moves your landed cost without any new tariff revision being published at all. If it touches a code you hold, you get one email naming the SKUs. It is not sold separately here.

Common questions

What counts as a classification?

One product getting an HS code from our AI. A product with several size or color variants counts once — classification happens at the product level, and the approved code applies to all of its variants.

Am I charged twice for the same product?

No. A product is only reclassified if you materially change it. That means its title, description, product type, vendor, tags or materials — the fields that can change which code is correct. A bulk edit across many products can therefore be a real billable event, so it is worth knowing before you run one. Re-syncing your catalog, re-opening the app, or approving a code costs nothing, and an unchanged product is never billed again.

What if I only want to try it?

Install it and classify up to 25 products a month without paying anything. There is no trial to expire and no card to cancel — if you stop using it, you stop being charged.

Is there a catalog size limit?

No. Large catalogs get the cheapest rates, and nothing is held back or put in a queue behind a plan limit.

How does this compare to buying prepaid credits?

The main alternative sells credits up front — as of July 2026, packs of $19 for 300, $59 for 1,000, $119 for 2,500, $199 for 6,000 — and keeps tariff-change monitoring in a separate $9/month add-on. The differences that hold at every catalog size are structural rather than about the sticker price. Here you pay for products actually classified: nothing to buy before you know it works, nothing to size in advance, nothing left over to expire, and after the first run only new or changed products are billed — so a catalog that is not growing is a bill that is not growing. Packs come in fixed sizes and catalogs do not, so a pack has to be rounded up to. A 400-product catalog is already past the 325 the smallest pack covers, so the cheapest way to cover it is 2 packs $38 — and 225 of those credits are never used. The same 400 products cost $23.25 here.

What the per-classification price pays for is the work described above: the second pass that picks the 10-digit line from the tariff lines that actually exist under the chosen subheading, the frontier model on the paid path, and the monitoring that is included here instead of being sold on its own. Those are costs we carry on every product classified, which is why the answer to “why not just be the cheapest” is that we are not trying to be.

On price alone it depends on your catalog, so here it is both ways. At 1,025 products we are $57.13 against $59 for the pack that covers it, and at 2,525 products we are $109.63 against $119. Around 325 products the two are within pennies of each other. Above that it stops being a straight line. They sell fixed-size packs, so whichever of us is cheaper flips back and forth depending on how neatly a pack fits your catalog — a catalog just over a pack boundary makes them buy a whole further pack, and one just under it wastes credits you paid for. Their $199 pack is the clearest case: it covers 6,025 classifications, and a catalog that size costs $229.05 here, so there they are ahead — and they stay ahead either side of that size too, over a stretch of catalog sizes in both directions, before the arithmetic flips back again. Which of us is cheaper depends on where your catalog happens to fall between their pack sizes, so the only comparison worth anything is the one at your own number — and prices move, theirs and ours, so check both before you decide.

How do I cancel?

Uninstall the app from your Shopify admin. Billing runs entirely through Shopify, and because there is no recurring fee, uninstalling means you are simply not charged again — there is no subscription to chase.

What happens to my HS codes if I leave?

They are yours. Approved codes live on your Shopify products, so they keep working after uninstall. We delete all of your shop’s data from our systems within 30 days of uninstall.

Do you touch my customers’ data?

No. DutyLookout reads product catalog data only — we never request access to customers or orders. See the privacy policy.

Try it on your real catalog.

Install, classify your first 25 products for free, and keep going only if it earns its keep.

Install on Shopify — no subscription